Cloud & Infrastructure
Cloud architecture for teams that cannot afford downtime
Resilience is a set of ordinary decisions made early: boundaries, budgets, backups and the discipline to test recovery.
Zen Tech Infrastructure · 8/20/2026 · 6 min read
Downtime is rarely caused by an exotic failure. It is caused by an ordinary failure meeting an untested assumption.
Design for the failure you expect
Regions go down. Dependencies rate limit. Certificates expire. Each of these has a known mitigation, and the cost of adding it before launch is a fraction of adding it during an incident.
Budgets are an architecture decision
Autoscaling without a ceiling turns an outage into an invoice. Set limits, alert on trajectory rather than thresholds, and make the owning team see the number weekly.
Backups you have not restored do not exist
Schedule a restore rehearsal the way you schedule a deployment. The first time you read a backup should never be the worst day of the quarter.
Observability before optimisation
You cannot tune what you cannot see. Traces, structured logs and per-tenant metrics turn an argument about performance into a measurement.